University of Sheffield Alcohol Minimum Price Modelling Research: Update

Transcription

1 MODEL-BASED APPRAISAL OF ALCOHOL MINIMUM PRICING AND OFF-LICENSED TRADE DISCOUNT BANS IN SCOTLAND USING THE SHEFFIELD ALCOHOL POLICY MODEL (v 2):- AN UPDATE BASED ON NEWLY AVAILABLE DATA APRIL 2010 ScHARR, University of Sheffield. 1

2 AUTHORSHIP Modelling team: Dr Yang Meng, Dr Robin Purshouse, Professor Alan Brennan Principal investigator: Dr Petra Meier CONFLICTS OF INTEREST The authors have no conflicts of interest. ACKNOWLEDGMENTS ScHARR would like to thank: Marjorie Marshall and colleagues at the Scottish Government for access to Nielsen market research data; Frank Dixon and colleagues at the General Register Office for Scotland for preparation of mortality data; Joan Corbett and colleagues at the Scottish Centre for Social Research for access to Scottish Health Survey data for 2008; Laura Wood and colleagues at NHS National Services Scotland for access to Scottish Schools Adolescent Lifestyle and Substance Use Survey data for 2008; and Stuart King at the Scottish Government for advice on data used in the model. We also thank Crispin Acton at the UK Department of Health for access to market research data procured from CGA Strategy. The original data creators, depositors or copyright holders, the funders of the Data Collections (where different) and the UK Data Archive bear no responsibility for the analysis and interpretation of the Expenditure and Food Survey, Labour Force Survey, and Offending and Justice Survey data sets downloaded from the Economic and Social Data Service and used in this study. 2

3 EXECUTIVE SUMMARY SCOPE OF RESEARCH In 2009, ScHARR developed an adaptation of the Sheffield Alcohol Policy Model for the population of Scotland. The findings of this modelling work Model-based appraisal of alcohol minimum pricing and off-licensed trade discount bans in Scotland A Scottish adaptation of the Sheffield Alcohol Policy Model version 2. Report to the Scottish Government. September was published on 28 September 2009 (Purshouse et al, 2009b). Since this time, a series of more up to date datasets have become available. New consumption data from the Scottish Health Survey (SHeS) and Scottish Schools Adolescent Lifestyle and Substance Use Survey (SALSUS) has become available for 2008 (the data in the previous model relates to 2003 for SHeS and 2006 for SALSUS). The Scottish Government has also procured market research data on the 2008/09 price distribution of off-trade alcohol (in terms of ethanol content) in Scotland from The Nielsen Company. New mortality data is available for Police recorded statistics and Scottish and Justice Survey (SCJS) data are now available for 2008/09. The requirement of this research project was to update the previous Scotland model with the new data to provide revised estimates of the effects of current proposals for minimum pricing and prohibition of off-trade discounting. The technical details of the methodology have not been changed. Therefore the methods section (Chapter 2) of the previously published report was not reproduced. The set of policies analysed remains the same as in the original research: 1. What are the likely effects of introducing a minimum unit price on alcohol consumption, sales, health, and workplace harms in Scotland? 2. What are the likely effects of introducing a ban on price-based promotions in the offlicensed trade in Scotland? 3. What are the likely effects of introducing a minimum unit price simultaneously with a ban on price-based promotions in the off-licensed trade in Scotland? 3

4 SUMMARY OF MODEL FINDINGS Overview M1. This updated report, which incorporates data on alcohol consumption data from the 2008 Scottish Health Survey (SHeS) and the 2008 Scottish Schools Adolescent Lifestyle and Substance Use Survey (SALSUS), market research data on the 2008/09 price distribution of off-trade alcohol, and updated mortality (2008) and (2008/09) statistics, shows results which are broadly similar to those presented in the September 2009 Sheffield report. M2. Increasing levels of minimum pricing show steep increases in effectiveness. Minimum Unit Price % change in consumption Minimum Price (only) Minimum Unit Price plus Discount Ban 25p -0.1% -3.2% 30p -0.3% -3.4% 35p -1.0% -4.0% 40p -2.3% -5.1% 45p -4.3% -6.7% 50p -6.7% -8.7% 55p -9.5% -11.2% 60p -12.3% -13.7% 65p -15.3% -16.5% 70p -18.4% -19.5% M3. A total ban on off-trade discounting is estimated to change overall consumption by -3.1%. Note that the total ban is assumed to prohibit all forms of price-based promotion, including straight discounting from list price in addition to multi-buy offers (such as buy three for the price of two ). Less restrictive types of ban have not been appraised because the market research data available to the study does not differentiate between types of price-based promotion. If the Scottish implementation of restrictions to off-trade discounting excludes particular types of discounting then the results reported here may overestimate the effectiveness of the policy. 4

5 M4. At lower minimum price thresholds, the combined effect of an off-trade discount ban and a minimum price is close to the individual effects of the two polices added together. At higher minimum price thresholds, the marginal increased effectiveness of an off-trade discount ban is reduced. Changes in levels of health, and workplace harm 1 M5. Low minimum price thresholds (e.g. 25p per unit) have little impact at reducing harmful outcomes. M6. As the minimum price threshold increases, alcohol-related hospital admissions and deaths are estimated to reduce. For example, a 30p threshold plus discount ban is estimated to reduce annual admissions by 3,200 at full effect, compared to 4,800 and 8,400 for 40p and 50p thresholds, respectively. The majority of health harm reductions are in chronic diseases. This is because much alcohol-attributable health harm occurs in middle/older age groups at risk of developing and potentially dying from chronic disease. Most of the prevented deaths occur in harmful drinkers. M7. As the minimum price threshold increases, alcohol-related s are estimated to reduce: e.g. -3,100 offences per annum for a 40p threshold (with discount ban) compared to -5,300 offences per annum for a 50p threshold (with discount ban). reductions take place across the spectrum of violent, criminal damage and acquisitive s. M8. -related harms are estimated to reduce proportionately less than health-related harms: e.g. for a 50p minimum price with discount ban, hospital admissions at full effect are estimated to reduce by 13.8% whilst s reduce by 3.0%. This occurs because the population sub-groups most associated with alcohol-related younger male drinkers consume a greater proportion of their alcohol in the on-trade sector when compared to the population average, and therefore are less affected by policies such as minimum pricing which impact more in the off-trade sector. M9. As the minimum price threshold increases, absenteeism from work is estimated to reduce: a minimum price of 40p with discount ban is estimated to reduce days absent 1 Note that the example policies shown here have been changed from the 2009 report for better consistency with latter sections of the report and figures reported in the media. All examples refer to minimum price policies with an off-trade discount ban (rather than minimum price policies alone). 5

6 from work by approximately 30,500 per annum, whereas for 50p the reduction is estimated at 51,200. M10. As the minimum price threshold increases, unemployment due to alcohol problems is estimated to reduce (in the model unemployment is a risk factor only for harmful drinkers). For a 40p threshold (with discount ban), 1,200 avoided cases of unemployment are estimated per annum; for 50p the figure is 2,000. Note that the estimated unemployment effects are based on evidence of association studies, rather than detailed prospective analysis of the dynamic effects of employed people becoming unemployed as a consequence of their drinking behaviour, or of unemployed people becoming employed again as a consequence of reductions in alcohol consumption. The estimated effects make no assumption about the direction of these processes and there is no analysis of how the current economic climate might affect these findings. Changes to consumer spending and retailer revenue M11. Consumer spending is estimated to increase under all policies. For example, under a combined 40p minimum price with discount ban policy, overall spending increases by 4.4%. M12. Retailer revenue from the sale of alcoholic beverages is estimated to increase under all policies: the model predicts increases in both off-trade and on-trade sectors (excluding duty and VAT). Higher minimum prices lead to greater additional retail revenues; however the model does not provide a breakdown of the revenue across the supply chain. For a 40p minimum price with combined discount ban, total retailer revenue is estimated to increase by 113m per annum. M13. Effects on VAT and duty receipts are estimated to be relatively small, due to the counter-balancing nature of the two taxes: duty is applied to the volume of sales (which is reducing overall) but the VAT is applied to the monetary of sales (which is increasing overall). For a 40p minimum price plus discount ban, total receipts are estimated to reduce by 10m. Valuation of harm reductions 6

7 M14. As the minimum price threshold increases, the financial of harm reductions increases 2 : the overall cumulative discounted financial of harm reduction over ten years is estimated at 824m for a 40p minimum price plus discount ban policy; this valuation nearly doubles for a 50p threshold ( 1.4bn). The valuation continues to increase steeply as the threshold is incremented. M15. The largest financially d component of harm reduction is the estimated impact on health-related quality of life: for example, just over half of the total 824m harm reduction in the 40p minimum price plus discount ban scenario is from the financial valuation of health quality adjusted life years (using 50,000 per ). M16. As the minimum price threshold increases, are reduced: for example, direct of reduce by approximately 22m over 10 years for a 40p minimum price plus discount ban compared to 38m for a 50p threshold. Similarly the of the loss of victim quality of life changes from around 17m to 28m (using 81,000 per ). Policy effects on different population sub-groups M17. Those who buy the most alcohol are the most affected in both absolute and relative terms: changes in spending affect mostly harmful drinkers, with hazardous drinkers somewhat affected and spending for moderate drinkers affected very little. For example, for a 40p minimum price in combination with an off-trade discount ban, extra spending per drinker per annum for moderate, hazardous and harmful drinkers is estimated at 10, 55 and 126 respectively (corresponding to an average additional spend per week of 0.19 for moderate drinkers, 1.06 for hazardous drinkers and 2.42 for harmful drinkers). M18. For all minimum price scenarios, with or without the presence of an off-trade discount ban, the majority of the health and healthcare benefits come from the harmful drinking group (e.g. 41% of the reduction in hospital admissions estimated for a 40p minimum price plus discount ban policy) even though these represent a small minority (6%) of all drinkers. 2 The financial valuation includes direct health and social care cost savings, direct savings to the criminal justice system and victims, a financial attribution to the savings in quality adjusted life years ( 50,000 for health-related s and 81,000 for -related S), and absence and unemployment savings based on average earnings. 7

8 M19. Reductions in are spread more evenly between the three drinker groups than for health-related outcomes. For example, for an off-trade discount ban, the reduction in volumes per annum is estimated to comprise 400 from moderate drinkers, 800 from hazardous drinkers and 800 from harmful drinkers. M20. The absolute contribution of the three drinker groups to reductions in days of absence is spread evenly across groups, e.g. a 40p minimum price plus off-trade discount ban has an estimated reduction in absenteeism of 9,300 days p.a. for moderate drinkers, 10,500 for hazardous drinkers and 10,300 for harmful drinkers. All reductions in unemployment arise from the harmful drinking group (since only this group is assumed in the model to be at risk of alcohol-attributable unemployment). M21. The majority of the estimated financial of harm reduction comes from the reduction in harms associated with harmful drinkers. Of the 824m harm reduction estimated for a 40p minimum price in combination with a discount ban, just over half ( 426m) is from harmful drinkers. Comparison of previous and updated modelling results M22. Compared to the SHeS 2003, the SHeS 2008 data suggest a slight reduction in mean alcohol consumption at the total population level. This is not the case for each age/sex group, with consumption for those under the age of 25, and especially young women, increasing significantly. At the beverage level, the consumption of beer/cider, wine and RTD decreased; while the consumption of spirits increased, especially for women. M23. The market research data available from Nielsen shows price distributions for alcohol which are marginally different from those estimated in our September 2009 report. Average prices paid are slightly higher, and the proportion of alcohol purchased which is bought inexpensively, for example at a price below 40p per unit, is slightly lower than in the September 2009 report. M24. The combined effect of the two points above is that the estimated effects of a particular policy are slightly smaller in this updated report both in terms of effects on alcohol consumption (e.g. 5.1% instead of 5.4% consumption reduction for a 40p minimum price combined with an off-trade discount ban) and the various measures of impact on alcohol-related harm. 8

9 TABLE OF CONTENTS AuthorShip... 2 Conflicts of interest... 2 Acknowledgments... 2 Executive summary... 3 Scope of research... 3 Summary of model findings... 4 Table of contents Introduction Background Research questions addressed Methods Prices Off-trade price distribution On-trade price distribution Off-trade price-based promotion Consumption Scottish Health Survey SALSUS Regression model linking mean consumption to peak consumption Health harms harms Workplace harms Sensitivity analysis Preferences for off-trade consumption Results Estimated impacts Example policy analysis: 40p minimum price (scenario 4) Example policy analysis: Off-trade discount ban (scenario 11) Example policy analysis: 40p minimum price combined with off-trade discount ban (scenario 15)

10 3.1.4 Summary tables of pricing policies Scotland population Comparison with previous results Summary tables for consumption analysis of pricing policies by population sub-group Summary tables for health, and employment harms by population sub-group Summary tables for financial of harm reductions by population sub-group Sensitivity analyses Probabilistic sensitivity analysis Moderate versus heavy drinkers Preferences for off-trade versus on-trade consumption Protective effects of alcohol for coronary heart disease Alternative AFs Summary of Results and Discussion Summary of model findings Limitations Limitations in the model of the relationship between price and consumption Limitations in the model of the relationship between consumption and harmful outcomes Areas for possible future research References Appendices

11 1 INTRODUCTION 1.1 BACKGROUND In 2009, ScHARR developed an adaptation of the Sheffield Alcohol Policy Model for the population of Scotland. This modelling work Model-based appraisal of alcohol minimum pricing and off-licensed trade discount bans in Scotland A Scottish adaptation of the Sheffield Alcohol Policy Model version 2. Report to the Scottish Government. September was published on 28 September 2009 (Purshouse et al, 2009b). Since this time, new consumption data from the Scottish Health Survey (SHeS) and Scottish Schools Adolescent Lifestyle and Substance Use Survey (SALSUS) has become available for 2008 (the data in the previous model relates to 2003 and 2006 respectively). The Scottish Government has also procured market research data on the 2008/09 price distribution of off-trade alcohol (in terms of ethanol) in Scotland from The Nielsen Company. New mortality data are available for 2008 and recorded statistics are also available for 2008/09. The first published results from the Scottish and Justice Survey (SCJS) 3 for 2008/09 are also now available. The requirement of the new research was to update the previous Scotland model with the new data to provide revised estimates of the effects of current proposals for minimum pricing and prohibition of off-trade discounting. 1.2 RESEARCH QUESTIONS ADDRESSED The set of policies analysed remains the same as in the original research: 1. What are the likely effects of introducing a minimum unit price on alcohol consumption, sales, health, and workplace harms in Scotland? 2. What are the likely effects of introducing a ban on price-based promotions in the offlicensed trade in Scotland? 3. What are the likely effects of introducing a minimum unit price simultaneously with a ban on price-based promotions in the off-licensed trade in Scotland? 3 The Scottish and Justice Survey (SCJS) is a large-scale continuous survey measuring adults' experience and perceptions of in Scotland. The 2008/09 report presents the results for the first full year of the survey. It represents a major shift in design, methodology and sample size from previous surveys. 11

12 2 METHODS This section outlines the changes to the methods from the original Scotland model. These relate to changes in the sources of raw data and subsequent intermediate methodologies to derive parameters for the model (see Table 2.1). Note that no structural changes have been made to the Scotland model. For details of the original research methodology, please refer to Purshouse et al (2009b). Changes have arisen in the following areas of modelling: Model area Raw data change Derived model parameters refresh only Derived model parameters change to methods Prices Consumption Health harms harms Workplace harms Table 2.1: Summary of modelling changes in this update 2.1 PRICES In the latest version of the model for England (Purshouse et al, 2009a), the price distributions for beer/cider, wine, spirit and RTD in both the off-trade and on-trade, were based on English purchasing data from the Expenditure and Food Survey (EFS) adjusted to match, at the total population level, England & Wales sales data from The Nielsen Company and England-only data from CGA Strategy (The Nielsen Company, 2008; CGA Strategy, 2009). In the previous Scottish adaptation of the model, sales data for Scotland was not available to the research team and so the price distributions in the model were based on the English distributions but adjusted, again at the total population level, for the differences between the EFS prices in England and Scotland (see Purshouse et al (2009b) for further details). Note that sales data is generally regarded as more accurate than purchasing data (which is older and based on self-reports), but is limited in that it provides no information on who purchases the alcohol (e.g. in terms of age, sex, or level of consumption). 12

13 Now, for the first time, sales data for the off-trade in Scotland has become available to the research team (The Nielsen Company, 2009). Therefore, in the model update, only Scotlandspecific data is now used for the off-trade sector Off-trade price distribution There are two main stages for incorporating the data into the model: pre-processing and calibration of EFS prices. These stages are described below Pre-processing The Nielsen price data is described across a set of 34 Product categories (e.g. dark rum, golden rum, etc) in terms of 15 price ranges (defined in terms of per unit). Both sales (in ) and sales volume (in litres of beverage) are provided. For modelling purposes, the distributions must be aggregated to the four beverages of beer/cider, wine, spirit and RTD. The distributions must also be expressed in terms of units. Therefore ABV estimates were requested from Nielsen for each Product category and used to convert the Product sales volumes to units of alcohol. Products are then summed to produce the four categories, as shown in Table 2.2. Model category Included Products ABV estimate Beer/cider Lager non-alcoholic / low-alcohol 0.5% Lager commodity 3.2% Lager standard 4.0% Lager premium 5.0% Lager superstrength 8.0% Ale non-alcoholic / low-alcohol 0.5% Ale commodity 3.2% Ale standard 4.0% Ale premium 5.0% Ale superstrength 8.0% Stout commodity 3.2% Stout standard 4.0% Stout premium 5.0% Stout superstrength 8.0% Cider white / strong 7.5% Cider other cider 4.5% Perry 7.5% Wine Fortified wine 17.0% Light wine 12.5% Sparkling wine 12.5% Champagne 12.5% Spirit Vodka 37.5% Blended whisky 40.0% Gin 38.0% 13

14 Cream liqueurs 20.0% French brandy 37.5% White rum 37.5% Imported whisky 40.0% Liqueurs 35.0% Malt whisky 40.0% Dark rum 40.0% Cognac 40.0% Golden rum 35.0% RTD RTD 4.5% Table 2.2: Matching of Nielsen Product categories to model categories Note that since a minimum price policy is a non-linear operator on the price distribution (i.e. some prices change, whilst others do not), the price per unit estimates that describe the distribution need to be accurate to avoid bias in the results. Ideally this would involve converting natural units of beverage to units of alcohol at the stock-keeping unit (SKU) level. However this was not achieved in the Nielsen data and so applying ABV estimates at the level of Nielsen Product categories is the next best option (the same approach was adopted for the England model) Calibration of prices As in all previous models (Brennan et al, 2008; Purshouse et al, 2009a; Purshouse et al, 2009b) the EFS is used as the basis of the price distribution because it provides information on purchasing preferences for different population subgroups (e.g. 18-to-24 year old male hazardous drinkers, or 55-to-64 year old female moderate drinkers). The same calibration method is used as in Purshouse et al (2009a): the population-level EFS cumulative price distribution is linearly interpolated so that it matches the known price points available from Nielsen (14 in this case). The maximum and minimum prices in the distribution remain unchanged since Nielsen does not provide these s. In the model, the distribution is then decomposed into its constituent population subgroups. Since the Scottish EFS sample size is small in relation to the overall EFS (around one tenth of the English sample size), issues with small samples occur, in which the number of data to describe the price distribution of some subgroups is insufficient. Using the previous model breakdown, the sample size is below 5 for 126 out of 384 subgroups (by sex, age group and drink type). In order to increase the sample size, the original eight age groups are merged into three, which are years (previously 16-17, and years) years (previously and years) and 55+ years (previously 55-64, and 75+ years). Note that the merging of age-groups is only applied to off-trade price distributions and does not affect on-trade price distributions and alcohol consumption. 14

15 Comparison of estimated distributions The off-trade price distributions used in the previous and the new model, and the distributions based on the raw EFS Scotland data are shown in Figure

16 Figure 2.1: Comparison of Scotland price distributions for off-trade beverages 16

17 2.1.2 On-trade price distribution No market research data has been made available for the distribution of on-trade prices. Therefore the four distributions from the previous model are used, but adjusted to 2009 terms using the high-level Nielsen data for Scotland shown in Table 2.3 (Scottish Government, 2010). The Nielsen data is preferred to ONS price adjustors in this case because the full disaggregation to modelled categories is available (note that the ONS figures are quite similar). Beverage type Mean 2008 price Mean 2009 price Adjustment factor ( per unit) ( per unit) Beer/cider % Wine % Spirit % RTD % Table 2.3: On-trade price adjustment factors Off-trade price-based promotion No market research data has been made available for the distribution of price-based promotions in supermarkets and off-licences. Therefore the same assumption is made as in the previous model: that the Scottish market is characterised by the same pattern of off-trade discounting as the English market, in terms of the cumulative price distribution, using Nielsen Grocery Multiples data for England & Wales. Since the Scottish off-trade price distribution has been updated, this requires a new mapping to the 10 price ranges available for the England & Wales data. The mapping, for each of beer/cider, wine, spirit and RTD, is shown in Appendix CONSUMPTION Two population surveys provide baseline levels of consumption for the model the Scottish Health Survey (SHeS) for subgroups aged 16 and over, and the Scottish Schools Adolescent Lifestyle and Substance Use Survey (SALSUS) for the 11 to 15 year old subgroups. The previous model used data from 2003 for SHeS and 2006 for SALSUS (for both, applying the latest ONS units conversion assumptions). However fresh data for 2008 is now available and has been used as the new baseline in the model. 17

18 2.2.1 Scottish Health Survey Figure 2.2 and 2.3 present the distributions of weekly and peak alcohol consumption for males and females in Scotland based on the latest consumption data from SHeS The 2008 age and gender-specific distributions of alcohol consumption for adults (18+ years) in Scotland are presented in Appendix 2. Figure 2.2: Distribution of the mean weekly intake among individuals aged 16 years old and over (SHeS 2008) 18

19 Figure 2.3: Distribution of peak intake (maximum units drunk heaviest drinking day) among individuals aged 16 years old and over (SHeS 2008) Figures 2.4 and 2.5 compare the average consumption in SHeS 2003 and 2008 by beverage type for men and women, respectively. Figures 2.6 and 2.7 compare the average consumption in SHeS 2003 and 2008 by age group for men and women, respectively. Compared to the SHeS 2003, the SHeS 2008 data show a general reduction of alcohol consumption at the population level. 4 At the beverage level, the consumption of beer/cider, wine and RTD decreased; while the consumption of spirits increased, especially for women. Despite the overall reduction of consumption at the population level, the consumption of those under the age of 25, and especially young women, has increased. This is important because the year old age group represents the age group with the heaviest alcohol consumption. 4 Market research Nielsen data shows a stable consumption pattern between 2005 and 2009 at the population level. 19

20 Figure 2.4: Comparison of average consumption over 4 beverage types for male (16+ years) between SHeS 2003 and 2008 Figure 2.5: Comparison of average consumption over four beverage types for female (16+ years) between SHeS 2003 and

21 Figure 2.6: Comparison of average consumption by age group for men (16+ years) between SHeS 2003 and 2008 Figure 2.7: Comparison of average consumption by age groups for women (16+ years) between SHeS 2003 and

22 2.2.2 SALSUS Figures 2.8 and 2.9 compare the average consumption between SALSUS 2006 and 2008 by beverage type for male and female pupils, respectively. These show that, for both genders, average consumption has dropped markedly across beverage types. Figure 2.8: Comparison of average consumption by beverage type for male pupils (aged years) between SALSUS 2006 and 2008 Figure 2.9: Comparison of average consumption by beverage type for female pupils (aged years) between SALSUS 2006 and

23 2.2.3 Regression model linking mean consumption to peak consumption Since the price elasticities used in the model relate a change in price to a change in mean consumption, further modelling is required to estimate the effects on peak daily consumption (the proxy measure for heavy episodic drinking). As described in Purshouse et al (2009b), this is achieved by estimating an average relationship between relative change in mean consumption and relative change in peak consumption at subgroup level. The same methodology is applied in the new model, but using the SHeS 2008 consumption data. The resulting model parameters are shown in Appendix HEALTH HARMS Mortality rates for 2008 are now available for the 47 conditions used in the model, and have been applied as the new baseline. Risk functions for all acute and wholly attributable chronic conditions were re-estimated. For partially attributable chronic conditions, the risk functions from the literature continue to be used. The parameter estimates are provided in Appendix 4. Note that fresh morbidity data was not available when this study was carried out and new estimates of alcohol-attributable fractions (AAFs) were already available for the 2009 model. Therefore assumptions from the previous model are re-used. 2.4 CRIME HARMS The availability of 2008/09 data for Scotland provides an opportunity to revise the modelling assumptions around the baseline volumes of alcohol-related : 1. Baseline recorded volumes In the previous model, Scottish categories were matched with English categories (Purshouse et al, 2009b). For consistency, the same matching is used in the new model, but the volumes were adjusted using the scaling factors derived from the latest ( ) and the previous ( ) police recorded volumes. 2. Multipliers In the previous model, the multipliers used to uplift the recorded volumes to actual volumes were based on the British Survey (BCS), due to the small sample size and large confidence intervals of the multipliers derived from the Scottish and Victimisation Survey (SCVS) However, the sample size has increased significantly in the latest SCJS (2008/09) and the confidence intervals on the estimated multipliers are much reduced. Therefore, the multipliers in the new model are estimated from the latest SCJS whenever possible. Some assumptions are necessary where multipliers are not available for categories in the model: serious assault and minor assault use the assault multiplier; both housebreaking categories use the housebreaking multiplier; fireraising uses the vandalism multiplier; other theft uses a weighted multiplier based on other 23

24 household theft, theft from the person and other personal theft ; shoplifting and total sexual offences continue to use the multiplier from the England model because relevant multipliers are not available from the SCJS. The baseline police recorded volumes and multipliers used in the new and the comparison with the previous model are shown in Table 2.4. Overall, using the new data results in lower baseline volumes (-6.5%). The baseline estimate for two large contributors to volume, shoplifting and other theft, increased by 10% and 48% respectively, the estimates for two other high-volume s, such as minor assault and vandalism, have decreased by 46% and 70%. Such shifts lead to changes in estimated savings in relation to criminal justice system and victimisation. Based on the updated baseline total volumes, Appendix 5 presents the breakdown of total estimated offences by age and gender in Scotland. The updated relative risk functions are provided in Appendix 6. 24

25 category Police recorded volumes multipliers volumes Previous model New model Difference (%) Previous model New model Previous model New model Difference (%) Serious assault, other non-sexual s of violence 6,398 6, % ,516 14, % Robbery 3,064 2, % ,337 6, % sexual offences 6,552 6, % ,070 32, % Housebreaking - Domestic dwelling 12,437 12, % ,361 18, % Housebreaking - Domestic non-dwelling & other 13,006 13, % ,313 20, % Theft from a motor vehicle 6,727 6, % ,836 15, % Theft of a motor vehicle 12,105 11, % ,526 12, % Shoplifting 29,186 32, % ,918,600 3,204, % Other theft 57,918 58, % , , % Fire-raising 4,616 4, % ,849 11, % Vandalism etc 109, , % , , % Minor assault 72,770 73, % , , % 334, , % 4,272,492 3,994, % Table 2.4: Baseline police recorded volumes and multipliers in the new model and the comparison with the previous model 25

26 2.5 WORKPLACE HARMS New raw data has not been extracted for absenteeism and employment. The changes here relate to the re-estimation of risk functions for these outcomes as a result of changes to the estimates of baseline mean and peak consumption levels. The refreshed parameters for the risk functions are shown in Appendix SENSITIVITY ANALYSIS The original analysis of pricing policies included a set of sensitivity analyses that attempted to account for the uncertainty in the representation of both current alcohol purchasing and consumption in Scotland and how price changes might influence consumer behaviour. Key uncertainties around the relationship between alcohol consumption and the population-level risk of coronary heart disease, and between alcohol consumption and population-level risk of were also explored. In this study, all previous sensitivity analyses are rerun using updated model inputs. The results are reported in Section Preferences for off-trade consumption The split of consumption between off-trade and on-trade for each subgroup in the model is based on purchasing data from the EFS. Alternative evidence for Scotland on the split from The Nielsen Company was originally made available to the research team for the year 2007/08 and new data for 2008/09 has also now been made available. An updated comparison with the EFS data is shown in Table 2.5. Beverage type Scotland preference for off-trade alcohol (via EFS) new model Scotland preference for off-trade alcohol (via Nielsen) previous model Scotland preference for off-trade alcohol (via Nielsen) new model Beer/cider 49.2% 45.6% 48.5% Wine 93.2% 81.6% 85.9% Spirit 88.1% 72.6% 76.0% RTD 50.4% 60.3% 67.2% 71.3% 63.3% 67.1% Table 2.5: Comparison of preferences for off-trade alcohol between EFS and Nielsen data sources The reason for the discrepancy between EFS and Nielsen off-trade preferences is not fully understood. The Nielsen data is based on a combination of census and survey, whilst the EFS data is based on a 14-day purchasing diary. One hypothesis is that the modelling assumption that two weeks purchasing in the EFS is equivalent to two weeks consumption 26

27 is not always appropriate for the off-trade sector; a second hypothesis could be that off-trade purchasing is recorded more accurately (e.g. via till receipts) than on-trade consumption (e.g. which is subject to memory recall). It is recognised that recall methods tend to underestimate actual consumption levels. The impact of using the alternative Nielsen evidence has been tested by proportionately adjusting all subgroup off-trade preferences to reflect the alternative overall preferences shown in Table

28 3 RESULTS This section contains updated model results for minimum price policies ranging from 25p to 70p per unit, an off-trade discount ban in isolation, and an off-trade discount ban in tandem with minimum price policies. Results are reported for Scotland as a whole and separately for moderate, hazardous and harmful drinkers. Note that moderate drinkers are defined as drinkers aged 18 years and over (reflecting the existing statutory minimum legal purchase age for alcohol), whilst other groups include persons aged 11 and over. 3.1 ESTIMATED IMPACTS Consumption and harm impacts across all policies are shown for the total population of Scotland in Table 3.7 to Table As in our previous report, the results of three example policies (40p minimum price, off-trade discount ban and the combination of the two policies) are presented in detail Example policy analysis: 40p minimum price (scenario 4) Table 3.1 shows the results for consumption changes, consumer spending and sales for a 40p minimum price policy. Table 3.2 shows the effects of the policy scenario on health, and workplace harms, as well as financial valuation. Overall weekly consumption changes by -2.3%. Consumption is estimated to reduce by 18.0 units per drinker per year. Consumption changes are greatest for harmful drinkers ( units per year). Moderate drinkers are affected in a small way in absolute terms (-2.9 units per year). Effects on health are estimated to be substantial with alcohol-attributable deaths estimated to reduce by approximately 26 within the first year of implementation and a full effect after 10 years of around 119. Deaths are distributed differentially across the groups, with approximately 6 saved amongst moderate drinkers, 51 amongst hazardous drinkers and 62 amongst harmful drinkers. Alcohol-attributable morbidity also decreases with an estimated reduction of 410 acute and 90 chronic es in year 1. Hospital admissions are estimated to reduce by around 640 in year 1, and a full effect after 10 years of 2,230 avoided admissions per annum. Healthcare service are estimated to reduce by 2.8m in year 1, with a gain d at 7.3m. is estimated to fall by 1,400 offences overall. Almost 75% of this annual reduction is amongst harmful drinkers, with approximately 20% amongst hazardous drinkers. The harm avoided in terms of victim quality of life is d at 0.7m per year. Direct of are estimated to reduce by 1.2m per year. 28

29 Workplace harms are estimated to be reduced by 730 fewer unemployed people and 11,800 fewer sick days per year. The societal of these harm reductions is estimated at 398m in total over the 10 year period modelled. In the first year, the estimated societal of the harm reductions is as follows: NHS cost reductions ( 2.8m), of s saved ( 7.3m), saved ( 1.2m), of s saved ( 0.7m) and employment related harms avoided ( 17.4m). The societal of harm reductions is distributed differentially across the groups, with hazardous drinkers accounting for 89m of the total, harmful drinkers 254m and moderate drinkers 52m. Absolute reductions in consumption are estimated to be largest in off-trade beer/cider and off-trade spirit. There is a substantial absolute increase in consumption of ontrade beer/cider. The cost impact of the policy on consumers varies substantially between drinker types: Harmful drinkers: 85 per drinker per annum Hazardous drinkers: 37 per drinker per annum Moderate drinkers: 5 per drinker per annum. An overall increased spend by consumers is estimated of 68m per annum, split roughly 63:37 between off-trade and on-trade sectors. Overall revenue to the Treasury (from duty and VAT receipts) changes by - 2.4m. 29

30 Males and Females Consumption Patterns Scotland Moderate Hazardous Harmful Baseline Mean consumption per person per week n people 4,527,300 3,062, , ,012 Mean consumption per drinker per week n drinkers 3,789,079 2,535, , ,012 % binge (>8 males, >6 females) 21.6% 12.1% 50.4% 70.0% Mean scale of binge if binge occurs (units) Volume sales Off-trade Beer (units per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD , ,746.0 Value sales Off-trade Beer ( per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD ,098 2,501 Absolute change Mean consumption per person per week Mean consumption per drinker per week % change in mean consumption -2.3% -0.9% -1.4% -4.8% Change in volume of consumption Off-trade Beer (units per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD Change in of Off-trade Beer purchases (sales) Wine ( per drinker per year) Spirit RTD On-trade Beer Wine Spirit RTD Effect of policy on "pocket" Off-trade Beer if drinkers did not change Wine consumption Spirit ( per drinker per year) RTD On-trade Beer Wine Spirit RTD change in retailer Off-trade m 43.1 m 7.9 m 17.4 m 17.4 received m On-trade m 25.0 m 5.5 m 12.4 m 6.9 (after VAT+Duty) m 68.1 m 13.4 m 29.9 m 24.3 Change in VAT Off-trade - m m m m 6.1 & Duty Received On-trade m 7.6 m 1.6 m 3.8 m m 2.4 m 0.5 m m 4.0 % change in spend / sales Off-trade +3.8% +3.1% +4.1% +4.0% On-trade +2.2% +1.3% +2.8% +2.8% +2.8% +1.8% +3.3% +3.4% Change Pop'n Spend Off-trade m 33.1 m 6.7 m 14.8 m 11.3 (Sales) On-trade m 32.5 m 7.1 m 16.2 m 9.0 m 65.7 m 13.8 m 31.0 m 20.3 Table 3.1: Consumption effects: Detailed results table for 40p minimum price (scenario 4) 30

31 Males and Females Harm Reductions Absolute change Scotland Moderate Hazardous Harmful Health Deaths Chronic Changes Acute in Illnesses Chronic Acute Admissions Chronic Acute s per annum Value of 'saved' s -7,323,321-1,905,341-2,370,488-2,960,188 Cost ( ) Chronic -503,216-61, , ,106 Acute -2,325, , , ,307 ( ) -2,829, , ,140-1,102,413 Health Deaths p.a. Chronic Changes Acute per annum in 0 Illnesses p.a. Chronic Acute , Admissions p.a. Chronic -1, ,021 Acute , ,187 s per annum Cost ( ) Chronic -5,070, ,603-1,713,419-2,763,495 Acute -2,519, , , ,336 ( ) -7,590,431-1,346,522-2,652,409-3,559,831 Cumulative Discounted s -3, ,365-1,628 Health Change Discounted Costs -43,489,461-8,927,166-15,033,564-19,271,005 over 10 yrs Valye of Discounted s -192,964,806-42,444,071-68,252,343-81,402,531 Value of Health Changes -236,454,268-51,371,237-83,285, ,673,537 Volume Violent Changes Damage per annum Theft/Oth , ,014 Cost ( ) Violent -527,121 23, , ,801 Damage -438,684 22,056-90, ,550 Theft/Oth -195,049-1,642-44, ,091 ( ) -1,160,854 43, , ,442 s Violent Damage Theft/Oth Value of 'saved' s -722,540 35, , ,008 Employment Volume days -11,761-2,333-2,926-6,318 Changes Unempl people per annum Cost ( ) -997, , , ,039 Unempl -16,420, ,420,093 ( ) -17,417, , ,316-16,948,132 Summary Health Costs ( ) -2,829, , ,140-1,102,413 Financial Value Costs ( ) -1,160,854 43, , ,442 Harm Reduction Employment Costs ( ) -17,417, , ,316-16,948,132 Direct Costs ( ) -21,407, ,873-1,425,877-18,916,987 Health s ( ) -7,323,321-1,905,341-2,370,488-2,960,188 s ( ) -722,540 35, , ,008 Societal Value ( ) -29,453,548-2,806,128-3,941,480-22,436,183 Cumul 10 year Health Costs ( ) -43,489,461-8,927,166-15,033,564-19,271,005 Summary Costs ( ) -9,654, ,321-2,032,750-7,205,853 Financial Value Employment Costs ( ) -144,856,215-1,803,985-2,056, ,950,921 Harm Reduction Direct Costs ( ) -198,000,045-10,366,830-19,123, ,427,780 Health s ( ) -192,964,806-42,444,071-68,252,343-81,402,531 s ( ) -6,663, ,568-1,338,275-5,155,278 Societal Value ( ) -397,628,250-52,487,334-88,713, ,985,589 Table 3.2: Harm effects: Detailed results table for 40p minimum price (scenario 4) 31

32 3.1.2 Example policy analysis: Off-trade discount ban (scenario 11) Table 3.3 shows the results for consumption changes, consumer spending and sales for the total off-trade discount ban policy. Table 3.4 shows the effects of the policy scenario on health, and workplace harms, as well as financial valuation. Overall weekly consumption changes by -3.1%. Consumption is estimated to reduce by 24.4 units per drinker per year. Consumption changes are greatest for harmful drinkers ( units per year). Moderate drinkers are affected in a small way in absolute terms (-6.5 units per year). Effects on health are estimated to be substantial with alcohol-attributable deaths estimated to reduce by approximately 37 within the first year of implementation and a full effect after 10 years of around 162. Deaths are distributed differentially across the groups, with approximately 12 saved amongst moderate drinkers, 96 amongst hazardous drinkers and 54 amongst harmful drinkers. Alcohol-attributable morbidity also decreases with an estimated reduction of 630 acute and 110 chronic es in year 1. Hospital admissions are estimated to reduce by around 940 in year 1, and a full effect after 10 years of 2,980 avoided admissions per annum. Healthcare service are estimated to reduce by 4.2m in year 1, with a gain d at 11.1m. is estimated to fall by 2,000 offences overall. Almost 40% of this annual reduction is amongst harmful drinkers, with approximately another 40% amongst hazardous drinkers. The harm avoided in terms of victim quality of life is d at 1.3m per year. Direct of are estimated to reduce by 1.8m per year. Workplace harms are estimated to be reduced by 620 fewer unemployed people and 20,500 fewer sick days per year. The societal of these harm reductions is estimated at 492m in total over the 10 year period modelled. In the first year, the estimated societal of the harm reductions is as follows: NHS cost reductions ( 4.2m), of s saved ( 11.1m), saved ( 1.8m), of s saved ( 1.3m) and employment related harms avoided ( 15.7m). The societal of harm reductions is distributed differentially across the groups, with hazardous drinkers accounting for 172m of the total, harmful drinkers 215m and moderate drinkers 101m. 32

33 Absolute reductions in consumption are estimated to be largest in off-trade wine and off-trade beer/cider, with a notable reduction also occurring in off-trade spirit. There is a substantial absolute increase in consumption of on-trade beer/cider. The cost impact of the policy on consumers varies substantially between drinker types: Harmful drinkers: 66 per drinker per annum Hazardous drinkers: 27 per drinker per annum Moderate drinkers: 5 per drinker per annum. An overall increased spend by consumers is estimated of 60m per annum, split roughly 81:19 between off-trade and on-trade sectors. Overall revenue to the Treasury (from duty and VAT receipts) changes by - 7m. 33

34 Males and Females Consumption Patterns Scotland Moderate Hazardous Harmful Baseline Mean consumption per person per week n people 4,527,300 3,062, , ,012 Mean consumption per drinker per week n drinkers 3,789,079 2,535, , ,012 % binge (>8 males, >6 females) 21.6% 12.1% 50.4% 70.0% Mean scale of binge if binge occurs (units) Volume sales Off-trade Beer (units per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD , ,746.0 Value sales Off-trade Beer ( per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD ,098 2,501 Absolute change Mean consumption per person per week Mean consumption per drinker per week % change in mean consumption -3.1% -2.1% -3.3% -3.9% Change in volume of consumption Off-trade Beer (units per drinker per year) Wine Spirit RTD On-trade Beer Wine Spirit RTD Change in of Off-trade Beer purchases (sales) Wine ( per drinker per year) Spirit RTD On-trade Beer Wine Spirit RTD Effect of policy on "pocket" Off-trade Beer if drinkers did not change Wine consumption Spirit ( per drinker per year) RTD On-trade Beer Wine Spirit RTD change in retailer Off-trade m 48.4 m 12.3 m 19.9 m 15.6 received m On-trade m 11.5 m 2.3 m 5.9 m 3.3 (after VAT+Duty) m 60.0 m 14.6 m 25.7 m 18.9 Change in VAT Off-trade - m m m m 4.2 & Duty Received On-trade m 3.5 m 0.7 m 1.8 m m m m m 3.1 % change in spend / sales Off-trade +4.4% +5.1% +4.2% +4.0% On-trade +1.0% +0.5% +1.3% +1.4% +2.3% +1.8% +2.4% +2.6% Change Pop'n Spend Off-trade m 37.9 m 10.9 m 14.9 m 11.4 (Sales) On-trade m 15.1 m 3.0 m 7.7 m 4.4 m 53.0 m 13.9 m 22.6 m 15.8 Table 3.3: Consumption effects: Detailed results table for off-trade discount ban (scenario 11) 34